Millennium Data Scientist Interview Guide

Millennium's Data Scientist interview, as three candidates describe it, begins with a timed online screen and moves into conversations about statistics and applied work. Expect a real coding screen up front. One candidate describes a HackerRank-style round mixing Python and SQL, where speed mattered. Another reports a 60-minute assessment with multiple-choice questions and three coding problems, including a longest common subsequence dynamic programming task. A third describes a HackerRank with four problems, mostly order book reconstruction plus a dynamic programming problem.

The technical rounds lean on practical statistics and data handling. One candidate was asked to implement OLS regression in NumPy without scikit-learn. In a later round, the same candidate discussed time series analysis in pandas, including how to structure and manipulate temporal data. Another had a Zoom round on linear algebra, probability, and statistics. It included an order book class to complete with submit and cancel methods, a question on the maximum of correlated random variables, and ridge regression properties. That candidate called the questions on the easier side.

Later conversations can be more behavioral. One candidate walked through a data project and then discussed their alpha research pipeline, including how they test ideas, which statistics they use, and how they think about overfitting. Another worked through a verbal case study with no live coding.

All three candidates report no offer, and their accounts stress speed and clean execution more than exotic theory. One reports hearing nothing for about a month after the second round, so applicants should be prepared for uneven communication.